Learning from Extrapolated Corrections
This addresses the challenge for robotics of learning from limited user feedback, though it appears incremental as it builds on existing correction-based learning methods.
The paper tackles the problem of enabling robots to learn cost functions from user corrections instead of full demonstrations, by framing it as online function approximation. Simulation results and a user study indicate that using function spaces with non-Euclidean norms better captures user intent in uncluttered environments, leading to more accurate cost functions and improved user perceptions.
Our goal is to enable robots to learn cost functions from user guidance. Often it is difficult or impossible for users to provide full demonstrations, so corrections have emerged as an easier guidance channel. However, when robots learn cost functions from corrections rather than demonstrations, they have to extrapolate a small amount of information -- the change of a waypoint along the way -- to the rest of the trajectory. We cast this extrapolation problem as online function approximation, which exposes different ways in which the robot can interpret what trajectory the person intended, depending on the function space used for the approximation. Our simulation results and user study suggest that using function spaces with non-Euclidean norms can better capture what users intend, particularly if environments are uncluttered. This, in turn, can lead to the robot learning a more accurate cost function and improves the user's subjective perceptions of the robot.